Evaluation of Soil Moisture-Based Satellite Precipitation Products over Semi-Arid Climatic Region
Abstract
:1. Introduction
2. Materials and Methods
2.1. Study Area
2.2. Datasets
2.3. Methods
3. Results
3.1. The Performance of Selected SPPs to Monitor Precipitation Spatial and Temporal Variability
3.2. Uncertainty Analysis of SPPs on Monthly Scale
3.3. Uncertainty Analysis of SPPs on Daily Scale
3.4. Uncertainty Analysis of SPPs at Seasonal Scale
3.5. Probability Density Function (PDF) and Performance Diagram of All SPPs to Track Precipitation Events
4. Discussion
5. Conclusions
- GPM-SM2Rain was proficient in assessing the spatial variability of precipitation over Pakistan. However, all other products were inappropriate with respect to spatial variation of precipitation over the study area, whereas the performance of SM2Rain-CCI is reasonable.
- The overall assessment of all selected SPPs was better on a monthly scale as compared to daily scales.
- The GPM-SM2Rain outperformed all other SPPs in terms of probability of detection on daily and seasonal scales.
- GPM-SM2Rain had good capturing capacities in all seasons, whereas the performance of all other SPPs is unsatisfactory in all seasons.
- In humid climatic regions, the SM2Rain performed most reliably with a value of CC >0.7.
- All SPPs performed better in capturing light precipitation events as indicated by the Probability Density Function.
- Based on the outcomes of the evaluation indices, the in-situ topographical and climatic conditions substantially influence the performance of SPPs.
Author Contributions
Funding
Institutional Review Board Statement
Informed Consent Statement
Data Availability Statement
Acknowledgments
Conflicts of Interest
Appendix A
Sr. No. | Station | Lat | Long | Ele. (m) |
---|---|---|---|---|
1 | Astore | 35.37 | 74.90 | 2168.0 |
2 | Balakot | 34.38 | 73.35 | 981.0 |
3 | Bunji | 35.67 | 74.63 | 1470.0 |
4 | Burzil | 34.91 | 75.09 | 4030.0 |
5 | Chillas | 35.42 | 74.10 | 1251.0 |
6 | Cherat | 33.82 | 71.55 | 1372.0 |
7 | Chitral | 35.85 | 71.83 | 1500.0 |
8 | Deosai | 35.10 | 75.60 | 3910.0 |
9 | Dir | 35.20 | 71.85 | 1370.0 |
10 | Drosh | 35.57 | 71.78 | 1465.0 |
11 | G-Dopata | 34.20 | 73.60 | 813.5 |
12 | Gilgit | 35.92 | 74.33 | 1457.2 |
13 | G-Khan | 33.25 | 73.62 | 457.0 |
14 | Gupis | 36.17 | 73.40 | 2156.0 |
15 | Jhelum | 32.93 | 73.73 | 287.2 |
16 | Hunza | 36.32 | 74.65 | 2374.0 |
17 | Hushy | 35.37 | 76.40 | 3010.0 |
18 | Kakul | 34.18 | 73.25 | 1309.0 |
19 | Kallar | 33.42 | 73.37 | 518.0 |
20 | Khot | 36.52 | 72.58 | 3505.0 |
21 | Khunjrab | 36.85 | 75.40 | 4730.0 |
22 | Kotli | 33.52 | 73.89 | 614.0 |
23 | Mangla | 33.13 | 73.63 | 305.0 |
24 | Murree | 33.92 | 73.38 | 2127.0 |
25 | Muzaffarabad | 34.40 | 73.50 | 702.0 |
26 | Naltar | 36.22 | 74.27 | 2810.0 |
27 | Naran | 34.90 | 73.65 | 2363.0 |
28 | Peshawar | 34.00 | 71.93 | 327 |
29 | Plandri | 33.70 | 73.70 | 1402.0 |
30 | Rama | 35.36 | 74.81 | 3040.0 |
31 | Ratu | 35.15 | 74.81 | 2920.0 |
32 | RawlaKot | 33.87 | 74.27 | 1677.0 |
33 | Shigar | 35.53 | 75.59 | 2470.0 |
34 | Skardu | 35.34 | 75.54 | 2316.5 |
35 | S-Sharif | 34.82 | 72.35 | 970.0 |
36 | Ushkore | 36.02 | 73.36 | 3353.0 |
37 | Yasin | 36.63 | 73.30 | 3353.0 |
38 | Zani Post | 36.28 | 72.15 | 3000.0 |
39 | Ziarat | 36.83 | 74.28 | 3669.0 |
40 | PBO. Chhor | 29.88 | 69.71 | 5 |
41 | PBO. Hyderabad | 25.38 | 61.8 | 28 |
42 | PBO. Jiwani | 25.06 | 61.8 | 56 |
43 | PBO. Jacobabad | 28.3 | 68.46 | 55 |
44 | M.O S K.A.P. | 24.9 | 67.13 | 22 |
45 | PBO. Nawabshah | 26.25 | 68.36 | 37 |
46 | PBO. Panjgur | 26.96 | 64.1 | 968 |
47 | PBO. Pasni | 25.26 | 63.48 | 9 |
48 | M.O. Badin | 24.63 | 68.9 | 9 |
49 | M.O. Gwadar | 25.13 | 62.33 | 29.86 |
50 | M.O. Larkana | 27.53 | 68.23 | 52.7 |
51 | M.O. Lasbella | 26.23 | 66.16 | 87 |
52 | M.O. Padidan | 26.85 | 68.13 | 46 |
53 | M.O. Rohri | 27.66 | 68.9 | 66 |
54 | A.M. Moen-jo-daro | 27.36 | 68.1 | 51.8 |
55 | A.M. Ormara | 25.2 | 64.66 | 2 |
56 | A.M. Turbat | 25.98 | 63.06 | 155 |
57 | A.M. Sukkur | 27.7 | 68.86 | 68.5 |
58 | P.B.O/AER Karachi | 24.9 | 66.93 | 22 |
59 | Marine Met. Kiamari. Karachi | 24.9 | 66.93 | 22 |
60 | M.O. Mithi | 24.75 | 69.8 | 30 |
61 | A.M. Tandojam | 25.66 | 68.71 | 19.5 |
62 | M.O Dadu | 26.71 | 67.78 | 38 |
63 | M.O Mirpur Khas | 25.51 | 69 | 15 |
64 | M.O. Thatta | 24.75 | 67.9 | |
65 | A.M. Uthal | 25.81 | 66.61 | 40 |
66 | A.M. Sakrand | 26.13 | 68.26 | 45 |
67 | Bahawal Nagar | 30 | 73.24 | |
68 | Bahawal Pur | 29.33 | 71.783 | |
69 | Bahawal Pur(A/P) | 29.383 | 71.683 | |
70 | Bhakkar | 31.616 | 71.06 | |
71 | Noorpur Thal | 31.866 | 71.9 | |
72 | Jauharabad | 32.5 | 72.43 | |
73 | Faisalabad | 31.43 | 73.13 | |
74 | Jhelum | 32.93 | 73.73 | |
75 | Khanpur | 28.65 | 70.683 | |
76 | Lahore A.P. | 31.583 | 74.4 | |
77 | Lahore PBO. | 31.55 | 74.33 | |
78 | Multan | 30.2 | 71.43 | |
79 | Mandi Bahauddin | 32.96 | 73.8 | |
80 | Sialkot | 32.516 | 74.53 | |
81 | Sialkot Airport | 32.53 | 74.03 | |
82 | Sargodha | 32.05 | 72.66 | |
83 | Toba Tek Singh | 30.983 | 72.783 | |
84 | D.G. Khan | 30.05 | 70.63 | |
85 | D.G. Khan (Aeromet) | 29.96 | 70.33 | |
86 | Jhang | 31.26 | 72.316 | |
87 | Mangla | 33.06 | 73.63 | |
88 | Sahiwal | 30.65 | 73.16 | |
89 | Chakwal | 32.916 | 72.85 | |
90 | Gujranwala | 32.36 | 74.35 | |
91 | Okara | 30.8 | 73.43 | |
92 | Rahim Yar Khan | 28.43 | 70.316 | |
93 | Gujrat | 32.56 | 74.06 | |
94 | MCC Lahore | 31.583 | 74.4 | |
95 | Rawalpindi | 33.56 | 73.02 |
References
- Li, S.; Jiang, C.; Ma, Y.; Long, Y.; Wu, R.; Zhu, Q.; Li, D.; Li, C.; Ning, Z. Adaptation Analysis in IMERG Precipitation Estimation for the Dongting Lake Basin, China. Atmosphere 2022, 13, 1735. [Google Scholar] [CrossRef]
- Nadeem, M.U.; Waheed, Z.; Ghaffar, A.M.; Javaid, M.M.; Hamza, A.; Ayub, Z.; Nawaz, M.A.; Waseem, W.; Hameed, M.F.; Zeeshan, A.; et al. Application of HEC-HMS for Flood Forecasting in Hazara Catchment Pakistan, South Asia. Int. J. Hydrol. 2022, 6, 7–12. [Google Scholar] [CrossRef]
- Anjum, M.N.; Irfan, M.; Waseem, M.; Leta, M.K.; Niazi, U.M.; Rahman, S.; Ghanim, A.; Mukhtar, M.A.; Nadeem, M.U. Assessment of PERSIANN-CCS, PERSIANN-CDR, SM2Rain-ASCAT, and CHIRPS-2.0 Precipitation Products over a Semi-Arid Subtropical Climatic Region. Water 2022, 14, 147. [Google Scholar] [CrossRef]
- Thaler, S.; Brocca, L.; Ciabatta, L.; Eitzinger, J.; Hahn, S.; Wagner, W. Effects of Different Spatial Precipitation Input Data on Crop Model Outputs under a Central European Climate. Atmosphere 2018, 9, 290. [Google Scholar] [CrossRef] [Green Version]
- Cheema, M.J.M.; Bastiaanssen, W.G.M. Local Calibration of Remotely Sensed Precipitation from the TRMM Satellite for Different Periods and Spatial Scales in the Indus Basin. Int. J. Remote Sens. 2012, 33, 2603–2627. [Google Scholar] [CrossRef]
- Palharini, R.; Vila, D.; Rodrigues, D.; Palharini, R.; Mattos, E.; Undurraga, E. Analysis of Extreme Precipitation and Natural Disasters Events Using Satellite Precipitation Products in Different Regions of Brazil. Atmosphere 2022, 13, 1680. [Google Scholar] [CrossRef]
- Pedreira, A.L.; Biudes, M.S.; Machado, N.G.; Vourlitis, G.L.; Geli, H.M.E.; Dos Santos, L.O.F.; Querino, C.A.S.; Ivo, I.O.; Neto, N.L. Assessment of Remote Sensing and Re-Analysis Estimates of Regional Precipitation over Mato Grosso, Brazil. Water 2021, 13, 333. [Google Scholar] [CrossRef]
- Bagtasa, G. Assessment of Tropical Cyclone Precipitation from GSMaP and GPM Products and Their Application to Analog Forecasting in the Philippines. Atmosphere 2022, 13, 1398. [Google Scholar] [CrossRef]
- Gebremichael, M.; Yue, H.; Nourani, V.; Damoah, R. The Skills of Medium-Range Precipitation Forecasts in the Senegal River Basin. Sustainability 2022, 14, 3349. [Google Scholar] [CrossRef]
- Abebe, S.A.; Qin, T.; Yan, D.; Gelaw, E.B.; Workneh, H.T.; Kun, W.; Shanshan, L.; Biqiong, D. Spatial and Temporal Evaluation of the Latest High-Resolution Precipitation Products over the Upper Blue Nile River Basin, Ethiopia. Water 2020, 12, 3072. [Google Scholar] [CrossRef]
- Cheema, M.J.M.; Bastiaanssen, W.G.M.; Rutten, M.M. Validation of Surface Soil Moisture from AMSR-E Using Auxiliary Spatial Data in the Transboundary Indus Basin. J. Hydrol. 2011, 405, 137–149. [Google Scholar] [CrossRef]
- Products, P. Quantitative Characteristics of the Current Multi-Source. Water 2021, 13, 334. [Google Scholar]
- Nwachukwu, P.N.; Satge, F.; El Yacoubi, S.; Pinel, S.; Bonnet, M.P. From Trmm to GPM: How Reliable Are Satellite-Based Precipitation Data across Nigeria? Remote Sens. 2020, 12, 3964. [Google Scholar] [CrossRef]
- Cattani, E.; Ferguglia, O. Precipitation Products’ Inter–Comparison over East And Southern Africa 1983–2017. Remote Sens. 2021, 13, 4419. [Google Scholar] [CrossRef]
- Bilal, H.; Govindan, R.; Al-Ansari, T. Investigation of Groundwater Depletion in the State of Qatar and Its Implication to Energy Water and Food Nexus. Water 2021, 13, 2464. [Google Scholar] [CrossRef]
- Amir, S.; Saqib, Z.; Khan, A.; Khan, M.I.; Khan, M.A.; Majid, A. Land Cover Mapping and Crop Phenology of Potohar Region, Punjab, Pakistan. Pakistan J. Agric. Sci. 2019, 56, 187–196. [Google Scholar] [CrossRef]
- Beck, H.E.; Vergopolan, N.; Pan, M.; Levizzani, V.; van Dijk, A.I.J.M.; Weedon, G.P.; Brocca, L.; Pappenberger, F.; Huffman, G.J.; Wood, E.F. Global-Scale Evaluation of 22 Precipitation Datasets Using Gauge Observations and Hydrological Modeling. Hydrol. Earth Syst. Sci. 2017, 21, 6201–6217. [Google Scholar] [CrossRef] [Green Version]
- Sharifi, E.; Eitzinger, J.; Dorigo, W. Performance of the State-of-the-Art Gridded Precipitation Products over Mountainous TerRain: A Regional Study over Austria. Remote Sens. 2019, 11, 2018. [Google Scholar] [CrossRef] [Green Version]
- Paredes-Trejo, F.; Barbosa, H.; dos Santos, C.A.C. Evaluation of the Performance of SM2Rain-Derived Precipitation Products over Brazil. Remote Sens. 2019, 11, 1113. [Google Scholar] [CrossRef] [Green Version]
- Wild, A.; Chua, Z.W.; Kuleshov, Y. Evaluation of Satellite Precipitation Estimates over the South West Pacific Region. Remote Sens. 2021, 13, 3929. [Google Scholar] [CrossRef]
- Hashim, M.; Reba, N.M.; Nadzri, M.I.; Pour, A.B.; Mahmud, M.R.; Yusoff, A.M.R.M.; Ali, M.I.; Jaw, S.W.; Hossain, M.S. Satellite-Based Run-Offmodel for Monitoring Drought in Peninsular Malaysia. Remote Sens. 2016, 8, 633. [Google Scholar] [CrossRef]
- Nadeem, M.U.; Anjum, M.N.; Afzal, A.; Azam, M.; Hussain, F.; Usman, M.; Javaid, M.M.; Mukhtar, M.A.; Majeed, F. Assessment of Multi-Satellite Precipitation Products over the Himalayan Mountains of Pakistan, South Asia. Sustainability 2022, 14, 8490. [Google Scholar] [CrossRef]
- Mu, N.; Rasool, G.; Ma, N.; Am, G.; Mm, J.; Hamza, A.; Aurangzaib, M.; Waheed, Z. Runoff Estimations by Using Satellite-Based Precipitation Products (SRPs) and Then Identifying Its Relation with Other Gauge Data Over Potohar Plateau, Pakistan. Int. J. Water Res. 2022, 4, 126. [Google Scholar]
- Hamza, A.; Anjum, M.N.; Cheema, M.J.M.; Chen, X.; Afzal, A.; Azam, M.; Shafi, M.K.; Gulakhmadov, A. Assessment of IMERG-V06, TRMM-3B42V7, SM2Rain-ASCAT, and PERSIANN-CDR Precipitation Products over the Hindu Kush Mountains of Pakistan, South Asia. Remote Sens. 2020, 12, 3871. [Google Scholar] [CrossRef]
- Masiza, W.; Chirima, J.G.; Hamandawana, H.; Kalumba, A.M.; Magagula, H.B. Do Satellite Data Correlate with In Situ Precipitation and Smallholder Crop Yields? Implications for Crop Insurance. Sustainability 2022, 14, 1670. [Google Scholar] [CrossRef]
- Al Abdouli, K.; Hussein, K.; Ghebreyesus, D.; Sharif, H.O. Coastal Runoff in the United Arab Emirates-the Hazard and Opportunity. Sustainability 2019, 11, 5406. [Google Scholar] [CrossRef] [Green Version]
- Zhang, H.L.; Wang, Y.J.; Wang, Y.Q.; Li, D.X.; Wang, X.K. Quantitative Comparison of Semi- and Fully-Distributed Hydrologic Models in Simulating Flood Hydrographs on a Mountain Watershed in Southwest China. J. Hydrodyn. 2013, 25, 877–885. [Google Scholar] [CrossRef]
- Oyerinde, G.T.; Fademi, I.O.; Denton, O.A. Modeling Runoff with Satellite-Based Precipitation Estimates in the Niger Basin. Cogent Food Agric. 2017, 3, 377–425. [Google Scholar] [CrossRef]
- Ahmad, M.J.; Iqbal, M.A.; Choi, K.S. Climate-Driven ConstRaints in Sustaining Future Wheat Yield and Water Productivity. Agric. Water Manag. 2020, 231, 105991. [Google Scholar] [CrossRef]
- Liu, Y.; Huang, Y.; Liu, Y.; Li, K.; Li, M. The Impact of Precipitation Movement Direction on Urban Runoff Cannot Be Ignored in Urban Hydrologic Management. Water 2021, 13, 2923. [Google Scholar] [CrossRef]
- Lu, X.; Tang, G.; Liu, X.; Wang, X.; Liu, Y.; Wei, M. The Potential and Uncertainty of Triple Collocation in Assessing Satellite Precipitation Products in Central Asia. Atmos. Res. 2021, 252, 105452. [Google Scholar] [CrossRef]
- Rahman, K.U.; Shang, S.; Shahid, M.; Wen, Y. Performance Assessment of SM2Rain-CCI and SM2Rain-ASCAT Precipitation Products over Pakistan. Remote Sens. 2019, 11, 2040. [Google Scholar] [CrossRef]
- Li, G.; Yu, Z.; Wang, W.; Ju, Q.; Chen, X. Analysis of the Spatial Distribution of Precipitation and Topography with GPM Data in the Tibetan Plateau. Atmos. Res. 2021, 247, 105259. [Google Scholar] [CrossRef]
- Ren, P.; Li, J.; Feng, P.; Guo, Y.; Ma, Q. Evaluation of Multiple Satellite Precipitation Products and Their Use in Hydrological Modelling over the Luanhe River Basin, China. Water 2018, 10, 677. [Google Scholar] [CrossRef] [Green Version]
- Anjum, M.N.; Ding, Y.; Shangguan, D.; Ahmad, I.; Ijaz, M.W.; Farid, H.U.; Yagoub, Y.E.; Zaman, M.; Adnan, M. Performance Evaluation of Latest Integrated Multi-Satellite Retrievals for Global Precipitation Measurement (IMERG) over the Northern Highlands of Pakistan. Atmos. Res. 2018, 205, 134–146. [Google Scholar] [CrossRef]
- Brocca, L.; Moramarco, T.; Melone, F.; Wagner, W. A New Method for Precipitation Estimation through Soil Moisture Observations. Geophys. Res. Lett. 2013, 40, 853–858. [Google Scholar] [CrossRef]
- Talchabhadel, R.; Nakagawa, H.; Kawaike, K.; Yamanoi, K.; Musumari, H.; Adhikari, T.R.; Prajapati, R. Appraising the Potential of Using Satellite-Based Precipitation Estimates for Evaluating Extreme Precipitation: A Case Study of August 2014 Event Across the West Rapti River Basin, Nepal. Earth Sp. Sci. 2021, 8, 1–15. [Google Scholar] [CrossRef]
- Li, L.; Hong, Y.; Wang, J.; Adler, R.F.; Policelli, F.S.; Habib, S.; Irwn, D.; Korme, T.; Okello, L. Evaluation of the Real-Time TRMM-Based Multi-Satellite Precipitation Analysis for an Operational Flood Prediction System in Nzoia Basin, Lake Victoria, Africa. Nat. Hazards 2009, 50, 109–123. [Google Scholar] [CrossRef]
- Andries, A.; Morse, S.; Murphy, R.J.; Lynch, J.; Woolliams, E.R. Assessing Education from Space: Using Satellite Earth Observation to Quantify Overcrowding in Primary Schools in Rural Areas of Nigeria. Sustainability 2022, 14, 1408. [Google Scholar] [CrossRef]
- Abd Elhamid, A.M.I.; Eltahan, A.M.H.; Mohamed, L.M.E.; Hamouda, I.A. Assessment of the Two Satellite-Based Precipitation Products TRMM and RFE Precipitation Records Using Ground Based Measurements. Alexandria Eng. J. 2020, 59, 1049–1058. [Google Scholar] [CrossRef]
- Bhati, D.S.; Dubey, S.K.; Sharma, D. Application of Satellite-Based and Observed Precipitation Datasets for Hydrological Simulation in the Upper Mahi River Basin of Rajasthan, India. Sustainability 2021, 13, 7560. [Google Scholar] [CrossRef]
- Nadeem, M.U.; Ghanim, A.A.J.; Anjum, M.N.; Shangguan, D.; Rasool, G.; Irfan, M.; Niazi, U.M.; Hassan, S. Multiscale Ground Validation of Satellite and Reanalysis Precipitation Products over Diverse Climatic and Topographic Conditions. Remote Sens. 2022, 14, 4680. [Google Scholar] [CrossRef]
- Habib, E.; Haile, A.T.; Sazib, N.; Zhang, Y.; Rientjes, T. Effect of Bias Correction of Satellite-Precipitation Estimates on Runoff Simulations at the Source of the Upper Blue Nile. Remote Sens. 2014, 6, 6688–6708. [Google Scholar] [CrossRef] [Green Version]
- Li, Z.; Wen, Y.; Schreier, M.; Behrangi, A.; Hong, Y.; Lambrigtsen, B. Advancing Satellite Precipitation Retrievals With Data Driven Approaches: Is Black Box Model Explainable? Earth Sp. Sci. 2021, 8, 1–15. [Google Scholar] [CrossRef]
- Huang, W.; Liu, P.; Chang, Y.; Liu, C. Evaluation and Application of Satellite Precipitation Products in Studying the Summer Precipitation Variations over Taiwan. Remote. Sens. 2020, 12, 347. [Google Scholar] [CrossRef]
Satellite Datasets | Spatial/Temporal Resolution | Time Coverage | Data Source |
---|---|---|---|
SM2Rain-ASCAT | 10 km\1-day | January 2007 to December 2021 | https://doi.org/10.5281/zenodo.2591214 |
GPM-SM2Rain | 0.25-degree\1-day | January 2007 to December 2018 | https://doi.org/10.5281/zenodo.3854817 |
SM2Rain-CCI | 0.25-degree\1-day | January 1998 to December 2015 | https://doi.org/10.5281/zenodo.1305021 |
SM2Rain | 0.25 degree/daily | January 1998 to December 2020 | https://explorer.adamplatform.eu/ |
Statistical Analysis | Details | Acceptable Range |
---|---|---|
CC = Correlation Coefficient | 1 | |
Gi = data of reference gauges | ||
G = average of the gauge data | ||
Ei = estimates of satellite/reanalysis product | ||
E = mean of the estimates of satellite/reanalysis product | ||
n = total number of datasets | ||
Ei = estimates of satellite/reanalysis product | 0 | |
Gi = data of reference gauges | ||
n = total number of datasets | ||
rbias = Bias, relative Bias | ±10 | |
Ei = estimates of satellite/reanalysis product | ||
Gi = data of reference gauges | ||
n = total number of datasets | ||
RMSE = Root Mean Square Error | 0 | |
Ei = estimates of satellite/reanalysis product | ||
Gi = data of reference gauges | ||
n = total number of datasets | ||
POD = Probability of Detection A = number of precipitation events that the SPPs/reanalysis products actually reported | 1 | |
B = number of precipitation events that the reference gauging stations observed but that the SPPs/reanalysis products missed | ||
FAR = False Alarm Ratio | 0 | |
C = number of precipitation events that the SPPs/reanalysis products misrepresented | ||
A = number of precipitation events that the SPPs/reanalysis products actually reported | ||
CSI = Critical Success Index | 1 | |
A = Amount of precipitation events that were really reported by SPPs and reanalysis products | ||
B = Amount of precipitation events missed by SPPs/reanalysis products while being observed by reference gauging stations | ||
C = Amount of precipitation events that were inaccurately represented by SPPs and reanalysis products |
SPPs | CC | BIAS (mm) | RMSE (mm) | rBIAS |
---|---|---|---|---|
GPM-SM2Rain | 0.91 | 5 | 10.13 | 150 |
SM2Rain-CCI | 0.75 | −0.3 | 9.45 | 180 |
SM2Rain | 0.81 | 4 | 8.75 | 200 |
SM2Rain-ASCAT | 0.82 | 3.88 | −0.4 | −100 |
SPPs | CC | BIAS (mm) | RMSE (mm) | rBIAS |
---|---|---|---|---|
GPM-SM2Rain | 0.39 | −0.2 | 7 | 100 |
SM2Rain-CCI | 0.21 | −0.1 | 18.75 | −80 |
SM2Rain | 0.35 | 0.2 | 34.98 | 110 |
SM2Rain-ASCAT | 0.22 | 0.4 | 13.24 | 140 |
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Asif, M.; Nadeem, M.U.; Anjum, M.N.; Ahmad, B.; Manuchekhr, G.; Umer, M.; Hamza, M.; Javaid, M.M.; Liu, T. Evaluation of Soil Moisture-Based Satellite Precipitation Products over Semi-Arid Climatic Region. Atmosphere 2023, 14, 8. https://doi.org/10.3390/atmos14010008
Asif M, Nadeem MU, Anjum MN, Ahmad B, Manuchekhr G, Umer M, Hamza M, Javaid MM, Liu T. Evaluation of Soil Moisture-Based Satellite Precipitation Products over Semi-Arid Climatic Region. Atmosphere. 2023; 14(1):8. https://doi.org/10.3390/atmos14010008
Chicago/Turabian StyleAsif, Muhammad, Muhammad Umer Nadeem, Muhammad Naveed Anjum, Bashir Ahmad, Gulakhmadov Manuchekhr, Muhammad Umer, Muhammad Hamza, Muhammad Mashood Javaid, and Tie Liu. 2023. "Evaluation of Soil Moisture-Based Satellite Precipitation Products over Semi-Arid Climatic Region" Atmosphere 14, no. 1: 8. https://doi.org/10.3390/atmos14010008
APA StyleAsif, M., Nadeem, M. U., Anjum, M. N., Ahmad, B., Manuchekhr, G., Umer, M., Hamza, M., Javaid, M. M., & Liu, T. (2023). Evaluation of Soil Moisture-Based Satellite Precipitation Products over Semi-Arid Climatic Region. Atmosphere, 14(1), 8. https://doi.org/10.3390/atmos14010008